Personal Website

Welcome! I am an Assistant Professor (tenure track) in econometrics at Erasmus University Rotterdam. Before that I was a Postdoc at Aarhus University. I obtained my PhD at the University of Amsterdam and the Tinbergen Institute.

My research advances the econometric methods in the field of causal inference with panel data. My current papers have applications in environmental economics, economic history and public health.

Master students at EUR under my supervision can schedule thesis meetings here.

---

Research

Fields: Econometrics, Causal inference, Applied microeconometrics

Working papers

1. Mediation Analysis in Difference-in-Differences Designs (honorable mention at the IAAE 2023)
   Abstract

   This paper develops strategies to understand the mechanisms behind treatment effects in difference-in-differences (DiD) designs. Building on concepts from mediation analysis, I present identification strategies for the part of the average treatment effect that is caused by the treatment affecting a mediating variable. The sequential DiD approach requires additional parallel trend assumptions, a restriction on the mediator effect heterogeneity, and monotonicity of the treatment effect on the mediator. To avoid some of these restrictions, I present a two-sample approach, which includes results from other studies. I propose robust inference procedures on the proportion of the total effect a particular channel can explain. I revisit two empirical studies to show how researchers can use these approaches in practice.

   
   

2. Time-Weighted Difference-in-Differences in Short Panels with Common Shocks (R&R Journal of Business & Economic Statistics) [previous version]
   Abstract

   This paper introduces a time-weighted difference-in-differences (TWDID) estimator for settings with few pre-treatment periods. Unlike conventional estimators, which use fixed pre-treatment weights, TWDID assigns variance-minimizing weights determined by the within-group covariance matrix of outcomes. The proposed estimator is efficient in the considered class when parallel trends hold across all periods. I introduce violations of parallel trends through common factors that have heterogeneous e!ects on the outcome. I show that the weights reduce the influence of the confounding factors, yielding a smaller bias than conventional DiD estimators under mild assumptions on the factors. Revisiting the impact of a cap-and-trade program on NOx emissions, TWDID yields smaller and more precise estimates than conventional approaches.

Work in progress

• Inference in Event Studies with Approximately Parallel Trends (with N. Harmon and M. Sølvsten)
• Policy Evaluation with Trending Data (with A. Pick)

---

Scheduled Talks

• Lund University (November 2025)
• University of Copenhagen (November 2025)
• EC-squared conference (December 2025)